[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-users] 'xl reboot domid' ignores all in-memory changes AND the domu's config file changes


  • To: xen-users@xxxxxxxxxxxxx
  • From: "Austin S. Hemmelgarn" <ahferroin7@xxxxxxxxx>
  • Date: Thu, 22 Sep 2016 07:28:27 -0400
  • Delivery-date: Thu, 22 Sep 2016 11:29:45 +0000
  • List-id: Xen user discussion <xen-users.lists.xen.org>

On 2016-09-21 15:33, Simon Hobson wrote:
Austin S. Hemmelgarn <ahferroin7@xxxxxxxxx> wrote:

A great description of self balloning and how to set it up. Not something I 
need, but very interesting to the geek in me ;-)

2. If you're doing this to overcommit memory (like I am), you need to have a 
very good understanding of your workload's memory requirements and how 
everything behaves, ...

I second this. it probably comes under the heading of "if you need to ask what the 
risks are, you shouldn't be doing it". Having seen the results of over-commit in 
disk storage and the hassles it causes, I don't think it's possible to stress too highly 
how big an issue it is when things ask for the resources you've promised them but don't 
have.
Fortunately, in the cases I've seen, it's been Windows HyperV and the 
hypervisor pauses all the VMs to prevent the train wreck that would otherwise 
happen. Still a PITA as the admin (not me !) has to find enough disk space to 
be able to unpause something and free up space within the virtual machine's 
disk - something of a catch 22 !
Yeah, although with memory there's better options to make sure you don't hit issues. In my case, I've got about 6 VM's on my home server using this, each providing a specific network service. Within each of them, everything that runs for more than a few seconds is confined in a Linux control group set to cap max memory usage. Because of this, any runaway programs will get caught with a localized OOM condition (within the cgroup instead of the system) before they can allocate enough to cause other issues. As a result of this, the only times the system might be vulnerable to memory exhaustion are when the VM's are booting (which I've got covered by serializing VM startup), and when I'm running updates on the VM's (which I also serialize).


Pretty much everything I know about it is stuff I've figured out from reading 
source code.  What I've got above is as close to a complete description as I've 
ever seen, maybe it should go on the Wiki?

I think it would be a very valuable wiki page.
I would think it should go on the transcendent memory page (which is in severe need of an update), or be linked from it, seeing as it's pretty much only possible because of transcendent memory support in Xen. I would put it up myself, but I don't have an account, and I really don't have the time to deal with getting one. It may also be useful to list what Linux distributions have kernels built with the required options. I can't help much with this though because I run custom kernel builds on all my systems.

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxx
https://lists.xen.org/xen-users

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.